Does diet breadth control herbivorous insect distribution size? Life history and resource outlets for specialist butterflies

Although butterfly distributions are known to be positively correlated with the number of larval hostplants used it is not known to what extent larval hostplant number uniquely influences butterfly distributions and to what extent effects are indirect through other variables. This issue is central to understanding the part generalism and specialism in host use play in organism persistence and conservation. Here, we have modelled the links between larval hostplant number and butterfly distributions using data from the UK. The model identifies the key variables that connect number of hostplants used by butterflies and the size of butterfly distributions. Significant correlations between variables give support to the model. Access to more hostplants is shown to affect a number of resource and life history variables impinging on butterfly population abundances and butterfly distributions. Butterfly distributions are largely accounted for (R^sup 2^>81%) by a set of resource and life history variables linked to numbers of hostplants: biotope occupancy, nectar sources used, utilities (the number of structures used by each life-cycle stage) and hostplant abundance. Application of partial regression demonstrates that the unique contribution of hostplant number to butterfly distributions is relatively small (R^sup 2^ = 14% to 33%), indicating that host use generalism has a limited direct impact on distributions. The modest correlations linking variables within the model illustrates that specialist phytophage feeders have a number of potential, distinct outlets, via resource and life history variables, to compensate for lack of supplementary larval hosts within their geographical ranges and enabling them to persist. Variables in the model each have considerable independence of action; without this, specialist feeders would have difficulty in expanding their distributions and acquiring new hosts, functionally-linked processes affecting evolutionary dynamics and persistence. We also question the nature of a direct functional link between local population abundance and distributions. Our model suggests a more complex functional relationship with implications for conserving insect herbivores.[PUBLICATION ABSTRACT]